This is a double-blind, placebo-controlled outpatient maintenance study examining the efficacy of fluoxetine in the relapse prevention of anorexia nervosa patients after their restoration to normal body weight. We propose this approach for several reasons. (1) Anorexia nervosa is associated with considerable morbidity and mortality and no treatment has been shown to prevent relapse. (2) Although inpatient care combining weight restoration and psychotherapy is usually effective in bringing about short-term recovery, the relapse rate within a year after discharge from inpatient care is about 60%. (3) Previous medication studies which have focused on the rate of weight gain as an index of efficacy, are of limited value because short-term rapid weight gain does not guarantee long-term recovery. Moreover, it is possible that the psychological and physiological disturbances present in emaciated patients may confound the effects of fluoxetine. We therefore consider that the most straightforward test of the efficacy of fluoxetine in anorexia nervosa is whether it will prevent weight loss in an outpatient setting where patients cannot be compelled to eat and maintain weight. Because anorectics use up a significant number of healthcare dollars, an effective outpatient treatment for this disorder would dramatically decrease the cost of treatment. We choose fluoxetine, a new serotonergic specific medication, as a treatment for anorexia nervosa, because it has been shown to be effective in disorders that often coexist in anorexia nervosa, such as depression, obsessive compulsive disorder, and bulimia nervosa. Most importantly, our open pilot study shows that only 22% of anorectics taking fluoxetine for a mean duration of 12 months have relapsed. Furthermore, in the preliminary (Lilly) phase of our double-blind trial, two patients (among the five who have participated in the study for at least 8 weeks) have relapsed and both were taking placebo. We will administer fluoxetine or placebo for 52 weeks to 80 female outpatients who have previously been restored to healthy weight in an inpatient eating disorders unit. A one year duration is chosen because favorable outcome at one year after discharge may predict a more benign eventual outcome. Patients will be stratified according to bulimic status and age of onset of illness because these factors may affect treatment response. Medication management and individual supportive psychotherapy will be delivered according to written manuals.